38 BUNSEN ON THE FORMATION OF 



tuent minerals separate first ; those which solidify less easily do 

 not separate until afterwards, in such a way as to render it un- 

 necessary to suppose this result to be aided by the power of 

 crystallization. It cannot therefore appear remarkable that an 

 originally homogeneous melted mass should yield a rocky mass, 

 richer in protoxide of iron and poorer in alumina at the com- 

 mencement of its solidification, and towards the end richer in 

 alumina and poorer in iron compounds. This may indeed be 

 easily ascertained to be the case by experiment. 



It is, in fact, by no means an uncommon circumstance in 

 Iceland, that the upper and lower ends of one and the same un- 

 decomposed trachytic column, situated in the midst of an exten- 

 sive mass of rock, present sensible diiFerences in composition. 

 The gray-coloured mass from the lower end of such a column in 

 the neighbourhood of Kalmanstunga had the composition 14, 

 and that from the opposite end of a white colour the composi- 

 tion 15 : — 



Silica .... 

 Alumina . . . 

 Protoxide of iron 

 Lime .... 

 Magnesia . . 

 Potash . . . 

 Soda .... 

 Water . . . 



14. 15. 



7274 71-49 



10-53 12-71 



6-57 4-27 



2-47 1-24 



1-61 0-08 



3-00 3-35 



2-33 3-73 



2-30 2-79 



101-45 99-66 



In like manner, the same mass of rock shows, even at distances 

 of a few feet, sensible differences in the relative quantities of 

 potash and soda. 



If these variations in composition are really the result of an 

 unequal separation of minerals, determined by conditions of tem- 

 perature and pressure, it is easy to understand that a decrease 

 in the quantity of one constituent must necessarily be attended 

 by an increase of the other, and that consequently the sum of 

 both can vary but little. All the above analyses prove this 

 to be really the case. The resemblance is restored when the 

 quantities of the especially-variable constituents are added 

 together : — 



